Continuing improvement in surgical technique, cardiopulmonary

Stroke After Coronary Artery Bypass Grafting Robert A. Baker, PhD, Lisa J. Hallsworth, BPsych(Hons), and John L. Knight, FRACS Cardiac Surgical Research Group, Cardiac and Thoracic Surgery, Flinders Medical Centre and Flinders University, Bedford Park, South Australia, Australia Background. Despite the continuing improvements in surgical and cardiopulmonary bypass techniques during cardiac surgery, stroke remains a devastating complication. This study aimed to identify the preoperative and intraoperative risk factors for developing a perioperative stroke in patients undergoing coronary artery bypass graft surgery on cardiopulmonary bypass. Methods. A total of 4,380 consecutive patients who received isolated coronary artery grafting on cardiopulmonary bypass between 1992 and 2002 were included. The sample contained three cardiopulmonary bypass temperature strategies: hypothermic (< 31 o C, n 1,853), tepid (32 35 o C, n 1,088), and normothermic (> 36 o C, n 1,439). Outcome measures reported include stroke incidence, 30-day mortality, and hospital length of stay. Results. The incidence of stroke was 1.2% (n 51). Stroke patients were older, were more likely to be diabetic, hypertensive, have creatinine levels greater than 0.12 mmol/l, and have a history of stroke than those who did not have stroke (p < 0.05). Multivariate logistic regressions identified diabetes (p 0.01), history of stroke (p 0.04), and older age (p 0.05) as independent predictors of stroke for all patients. The 30-day mortality for stroke patients was ten times greater than that of those who did not suffer stroke (17.6 vs 1.7%). Conclusions. Diabetes, history of stroke, and older age were identified as risk factors for stroke after coronary bypass; the temperature at which cardiopulmonary bypass was performed was not significant. (Ann Thorac Surg 2005;80:1746 50) 2005 by The Society of Thoracic Surgeons Continuing improvement in surgical technique, cardiopulmonary bypass, and anesthetic management have reduced the overall morbidity associated with cardiac surgery, and widened the spectrum of individuals being considered for surgery. Despite advances, central nervous system (CNS) injury remains a devastating and debilitating complication for many who undergo cardiac surgery, with the incidence of stroke reported to be approximately 2% to 5% after coronary artery bypass graft (CABG) surgery [1 5]. Multiple risk factors have been associated with postoperative stroke, including length of time on cardiopulmonary bypass (CPB), increasing age, male gender, history of diabetes mellitus, or hypertension [1, 6]. The role of systemic temperature management during CPB in influencing the etiology of stroke remains unclear in spite of a number of studies reporting the effect of normothermic bypass on patients undergoing coronary revascularization [7 11]. The low incidence of post-cabg stroke demands a large sample size to identify risk factors. The present paper takes advantage of data routinely collected for all CABG cases at two South Australian institutions, performed by three cardiac surgeons. Using a large database, collected prospectively, this study s retrospective analysis allows identification of potential risk factors for stroke. Accepted for publication April 28, 2005. Address correspondence to Dr Baker, Cardiac Surgical Research Unit, Department of Cardiac and Thoracic Surgery, Level 6, Flinders Private Hospital, Flinders Drive, Bedford Park, South Australia 5042; e-mail: rob.baker@flinders.edu.au. Material and Methods Patients who underwent an isolated CABG procedure utilizing CPB at either Flinders Medical Centre between February 1992 and February 2002, or Ashford Community Hospital between January 1995 and February 2002, were considered for inclusion in the study sample (n 4,511). Patients were excluded from analyses if there were incomplete data available for CPB temperature, postoperative stroke status, and/or postoperative outcome (ie, death within 30 days). Of these 4,511 patients, 4,380 (97%) were eligible for inclusion in analyses. The study was conducted under the auspice of the Clinical Governance Committee (Flinders Medical Centre). Additional patient consent was not required for access to deidentified, routinely collected data. Comparisons between the final sample of 4,380 patients and the 131 not eligible for inclusion due to missing data were conducted by a 2 test for categoric variables and an independent sample t test for continuous variables (Table 1). The excluded patient sample consisted of significantly less patients with Canadian Cardiovascular Society class 3 or 4 ischemic heart disease ( 2 6.0, p 0.05), more redo CABG operations ( 2 7.4, p 0.05), and preoperative creatinine levels greater than 0.12 mmol/l ( 2 12.4, p 0.05); other variables including age, sex, diabetes, hypertension, hypercholesterolemia, unstable angina, and previous stroke showed no differences. The primary outcome measure was an in-hospital perioperative or postoperative stroke. Stroke was defined based on the Society of Thoracic Surgeons definition as a focal neurologic deficit persisting for more than 72 hours [12]. All diagnoses were made by surgical staff, confirmed by either 2005 by The Society of Thoracic Surgeons 0003-4975/05/$30.00 Published by Elsevier Inc doi:10.1016/j.athoracsur.2005.04.059

Ann Thorac Surg BAKER ET AL 2005;80:1746 50 STROKE AFTER CABG Table 1. Preoperative, Intraoperative, and Postoperative Variables for All Patients Total n 4380 Age (yrs) 65.5 10 Age (% 70yr) 40.8 Sex (% male) 76.1 Weight (kg) 79.9 15.3 Diabetes (%) 24.8 Hypertension (%) 52.1 Hypercholesterolemia (%) 32.7 Unstable angina (%) 17.9 CCS 3 or 4 (%) 50.9 Elective (%) 76.0 Redo operation (%) 7.3 Previous stroke (%) 5.0 Renal failure (% Cr 0.12) 15.9 Preoperative AF (%) 2.9 LV dysfunction (% poor) 9.8 Number of grafts 2.6 0.8 Procedure time (min) 131.4 34.3 Bypass time (min) 48.1 22.7 X-clamp time (min) 28.3 11.1 ICU hours 45.5 77.8 Intub. time (min) 21.5 50.6 MI (%) 2.6 Frank psychosis (%) 1.5 AF (%) 86.4 Stroke (%) 1.2 Length of stay (day) 8.9 5.6 30-day mortality (%) 1.9 Data are presented as mean SD unless stated otherwise. AF atrial fibrillation; CCS Canadian Cardiovascular Society; Cr creatinine; ICU intensive care unit; LV left ventricular; MI myocardial infarction; X-clamp cross clamp. a neurologist and/or by brain computed tomographic scans, and were reviewed by research staff prior to data analysis. Surgical Procedure and Cardiopulmonary Bypass Strategies Cardiopulmonary bypass was undertaken using a standardized extracorporeal circulation polyvinyl chloride circuit (Cobe Cardiovascular, Arvada, CO) utilizing nonpulsatile flow, arterial line filtration (40 m; Medtronic, Anaheim, CA) and an alpha stat management protocol. Bypass was established by proximal aortic cannulation and single dual stage cannulation of the right atrium, with a membrane plate oxygenator (Cobe CML) or a hollow fiber membrane oxygenator (Maxima; Medtronic), or Capiox SX 18 or SX 25 (Terumo Corp, Tokyo, Japan). Three systemic perfusion strategies were performed during this period, 1,853 patients had their operation performed using hypothermic bypass strategy of systemic cooling to less than 32 C, 1,088 patients had tepid bypass (33 35 C), and 1,439 underwent normothermic 1747 systemic perfusion (systemic temperature more than 36 C). Active cooling was employed for all hypothermic cases where a target temperature of 32 C was required. Rewarming for all cases followed standard institutional guidelines, with the temperature gradient between blood and heat exchanger not exceeding 10 C, a maximum heat exchanger temperature of 41 C to avoid damage to the blood elements, and the rate at which the patient was rewarmed not exceeding 0.5 C per minute. Patients were separated from bypass with nasopharyngeal temperature in excess of 36.5 C. During this period both fibrillating heart and cardioplegic techniques were utilized. Three anterograde cardioplegic protection techniques were employed; cold blood cardioplegia (4 8 C) was used in patients undergoing hypothermic systemic perfusion, tepid blood cardioplegia (33 34 C) was used in patients undergoing tepid systemic perfusion, and warm blood cardioplegia ( 36 C) was used in patients undergoing hypothermic systemic perfusion. Initial cardioplegia utilized high potassium solutions (30 mmol/l of KCl) to achieve asystole with further boluses of a low potassium solution (15 mmol/l of KCl) to maintain asystole. Distal anastomoses were performed with the aortic cross clamp in place for all procedures using cardioplegia, no aortic cross clamp was used in fibrillating cases. All proximal anastomoses were performed with the aid of an aortic side-biting clamp in place. All patients received a standard moderate fentanylbased anesthetic technique and a standard heparinization protocol of 300 IU/kg heparin bolus prior to cannulation, with an additional 10,000 IU in the pump prime. Target activated clotting time greater than 400 was maintained throughout bypass. No patients received epiaortic scanning, and very few had intraoperative transesophageal echocardiography. Statistical Analyses All analyses were undertaken using the SPSS software package (SPSS Inc, Chicago, IL). An value of 0.05 or less was considered statistically significant, with the Bonferroni correction for multiple comparisons ( /number of comparisons) applied where appropriate. Patients were excluded list-wise in the case of incomplete data on any of the variables investigated. To identify variables significantly predictive of post- CABG stroke incidence, univariate logistic regression analyses were run for each preoperative variable investigated. Any variables that carried a significance of less than or equal to 0.20 alone were entered into a multivariate logistic regression together to identify significant predictors once other predictors were controlled for. Any variables entered into this multivariate logistic regression, which correlated highly with other variables in the analysis (ie, representative of the same measure/ construct), were then excluded and the regression reported independently to identify the influence of this correlation. Where variables violated the assumption of normality, appropriate transformations were made. If these trans- CARDIOVASCULAR

1748 BAKER ET AL Ann Thorac Surg STROKE AFTER CABG 2005;80:1746 50 formations did not prove useful, variables were transformed to categorical variables, where possible, or raw scores were used. Results Preoperative Variables Fifty-one out of the 4,380 patients were diagnosed with having intraoperative or postoperative stroke (1.2%). Preoperative, intraoperative, and postoperative data for the entire sample are summarized in Table 1. Table 3. Predictors of Postoperative Stroke Predictors Entered Odds Ratio CI p All Patients Diabetes 3.5 1.4 8.7 0.01 a Previous Stroke 6.3 1.1 36.6 0.04 a Age 1.1 1.0 1.1 0.05 a Hypertension 2.9 0.9 8.9 0.07 CVD 0.8 0.2 4.2 0.79 Cr 0.12 0.9 0.3 2.5 0.80 Preoperative AF 1.2 0.1 9.9 0.86 a p 0.05. Table 2. Preoperative, Intraoperative, and Postoperative Data for Postoperative Stroke Patients and No Stroke Patients Stroke n 51 No Stroke n 4329 Preoperative variables Age (yrs) 70.6 6.5 65.5 10.0.000 Age (% 70yr) 58.8 40.6.010 Sex (% male) 68.6 76.2 n/s Weight (kg) 74.2 13.5 80.0 15.3.008 Diabetes (%) 51 24.5.000 Hypertension (%) 76.6 51.9.001 Hypercholesterolemia (%) 31.0 32.7 n/s Unstable angina (%) 27.5 17.8 n/s CCS 3 or 4 (%) 52.9 50.9 n/s Elective (%) 68.6 76.1 n/s Redo (%) 9.8 7.4 n/s Previous stroke (%) 25.0 4.8.000 Renal failure (% Cr 32.7 15.7.003 0.12) Preoperative AF (%) 8.0 2.9 n/s LV dysfunction (% poor) 16.2 9.8 n/s Intraoperative variables Number of grafts 2.6 0.7 2.6 0.8 n/s Procedure time (min) 132.1 30.7 131.4 34.3 n/s Bypass time (min) 51.3 15.9 48.1 22.8 n/s X-clamp time (min) 26.2 9.2 28.3 11.1 n/s Minimum 32.7 2.7 32.9 2.2 n/s nasopharyngeal temperature ( C) Perfusion strategy 39.2/23.5/37.3 42.2/24.9/32.9 n/s hypothermic/tepid/ normothermic (%) Postoperative variables ICU hours 180.4 259.0 43.9 71.6.001 Intub. time (min) 16.4 10.2 21.5 50.9 n/s MI (%) 2.0 2.6 n/s Frank psychosis (%) 11.5 1.3.001 AF (%) 21.8 31.4 n/s Length of stay (day) 20.2 12.8 8.7 5.3.000 30-day mortality (%) 17.6 1.7.000 Data are presented as mean SD unless indicated otherwise. AF atrial fibrillation; Cr creatinine; ICU intensive care unit; LV left ventricular; MI myocardial infarction; X- clamp cross clamp. p AF atrial fibrillation; CI confidence interval; Cr creatinine; CVD cardiovascular disease. Stroke Versus Nonstroke Comparison of patients who suffered a postoperative stroke during surgical admission with those who did not is summarized in Table 2. Preoperatively, stroke patients were older (t 3.6, p.05), weighed less (t 2.7, p 0.05), were more likely to be diabetic ( 2 19.0, p 0.05), hypertensive ( 2 11.4, p 0.05), have creatinine levels greater than 0.12 mmol/l ( 2 10.4, p 0.05), and have a history of stroke ( 2 30.3, p 0.05) than those who did not suffer a postoperative stroke. Postoperatively, stroke patients spent more than four times as long in intensive care (t 12.4, p 0.05), twice as long in hospital overall (t 14.6, p 0.05), and were 10 times more likely to die within 30 days of operation ( 2 67.9, p 0.05). Logistic Regression Considering all patients together, seven preoperative variables were predictors of postoperative stroke alone (Table 3). When entered into a multiple logistic regression age at operation, diabetes and previous stroke were significant predictors of post-cabg stroke. When cerebrovascular disease was removed from the regression due to strong correlations with other predictors (eg, history of stroke [r 0.56, p 0.01]), no changes were found. Temperature management on CPB was not a predictor of stroke. Comment Stroke is a devastating and debilitating complication of cardiac surgery, increasing the risk of mortality and decreasing postoperative quality of life. Previous studies have found the incidence of stroke to be 2% to 5% after CABG surgery [1 5]. The etiology of stroke is thought to be multifactorial. This study set out to identify perioperative risk factors associated with stroke after CABG, including the influence of temperature on stroke and to examine the influence of outcome. Among the current study population of 4,380 consecutive CABG surgery patients, the stroke incidence was 1.2%. When patients who suffered postoperative stroke were compared with those who did not, they were more likely to be older, diabetic, hypertensive individuals who had previously

Ann Thorac Surg BAKER ET AL 2005;80:1746 50 STROKE AFTER CABG suffered a stroke and had high creatinine levels. The bypass management strategy and minimum nasopharyngeal temperature did not significantly influence the incidence of stroke in this sample of patients. These patients spent significantly longer in intensive care and hospital, and had a 30-day mortality of 17.6% compared with 1.7% of nonstroke patients. These differences are striking and illustrate the importance of identifying risk factors and finding mechanisms to prevent CNS injury from occurring. Of the variables that differed between patients that did and did not suffer stroke, the current study identified three independent preoperative risk factors of stroke using multiple logistic regressions: older age, diabetes, and previous history of stroke. Age has been identified in the majority of previous research as a risk factor of postoperative stroke [4, 13 15]. In the present study, patients who suffered a postoperative stroke had a mean age of 70.6 years compared with 65.5 for nonstroke patients. The effect of age is likely to be that of a moderator as it is associated with many other risk factors, including an increased incidence of carotid and aortic disease, systemic atherosclerosis, cerebrovascular disease [16], and cognitive decline [17]. A history of diabetes mellitus was a second independent predictor of stroke in our analysis, and has also been identified as a risk factor by other investigators [1, 15]. Fifty-one percent of stroke patients had diabetes mellitus compared with 24% of those without stroke. The potential mechanisms involved include the increased risk patients with diabetes have of atherosclerotic embolization and the impaired autoregulation of cerebral blood flow [1, 4, 6, 14]. Additionally, hyperglycemia may contribute to increased risk of adverse postoperative outcomes; hyperglycemia is associated with a worsening of postischemic brain injury [18]. The current study also identified history of stroke as a strong predictor of postoperative stroke, and this has been supported in most previous analyses [1, 4, 6, 13, 16]. Twenty-five percent of stroke patients had a history of stroke in the current study, compared with 4.8% of nonstroke patients. A history of stroke, or cerebrovascular disease in general, is indicative of pathologic conditions within the cerebrovascular system or carotid artery stenosis [13]. Preoperative hypertension, a commonly reported risk factor due to its potential for precipitating ischemic stroke, did not reach statistical significance for being predictive of stroke (p 0.07) in the entire group. Other risk factors previously reported that were not found to be independent predictors in our analysis included unstable angina, male gender, previous cardiac surgery, and urgency of operation [1, 6, 16]. This study did not identify the bypass management strategy, or minimum nasopharyngeal temperature on bypass, as risk factors for stroke. This is consistent with the majority of previous findings that compared hypothermic with normothermic techniques [7 9]. In contrast, two studies [10, 11] found a significant difference between the normothermic and hypothermic groups, reporting an increased risk of stroke for normothermic patients (3.1% 1749 vs 1.0% hypothermic). This is thought to be a direct result of normothermic perfusion reducing the brain s tolerance to cerebral hypoxia [11, 19, 20]. Further studies failed to replicate these results, with no significant differences reported [7 9] between normothermic and hypothermic techniques in stroke incidence. One of the reasons hypothesized for these different findings is the actual technique of normothermic bypass, with Martin and colleagues [10] utilizing greater than 36 o C perfusate bypass, while other groups that found no significant differences used lower perfusate temperature management. The present report also reports on patients who have undergone bypass utilizing a tepid temperature management strategy, in which patients are allowed to drift to the target temperature and then rewarmed. The lack of significant difference is in direct contrast with the findings of Martin and colleagues [10], which raises the question as to what role, if any, temperature management has in the etiology of postoperative stroke. However, definitive conclusions should not be drawn on the findings of this retrospective analysis. Limitations and Future Directions One limitation of this study is that it reports on patients operated between 1992 and 2002. During this time the influence of changing surgical methods must be considered, for example, the use of fibrillation techniques in a number of patients. Similarly the determination of systemic bypass temperature strategy was not independent of the surgeon, but determined by surgeon preference. A second limitation was the lack of thorough investigation into the impact of carotid bruits and/or stenosis on the outcome. Previous research has identified both bruits and stenosis as significant predictive factors for post- CABG stroke [21, 22]. The most likely mechanism behind carotid stenosis contributing to postoperative stroke is embolism from the atherosclerotic lesion of the carotid artery [22]. Naylor and colleagues [16], in a systematic review of the literature, identified a preoperative prevalence rate of carotid stenosis of 9.9%, leading to an almost fourfold excess risk of operative stroke for those with carotid bruits compared with those without (5.6% vs 1.6%). In the present paper, all patients who suffered strokes were checked retrospectively to determine the incidence of carotid stenosis or bruits. Of the 51 stroke patients, 9 patients (17.6%) had carotid bruits or carotid stenosis greater than 50%. We do not have complete carotid disease data for those who did not have strokes, and therefore a detailed analysis was not possible. The presence of significant carotid disease has been clearly associated with increased stroke; however, the management of patients with carotid disease and requiring bypass graft surgery clearly still provides a major management dilemma [21]. In conclusion, stroke causes permanent and irreversible neurologic damage, and as such, places huge constraints on a patient s quality of life and cognitive and physical abilities, as well as placing a financial and resource burden on health care systems. As such, the identification of methods by which perioperative CNS CARDIOVASCULAR

1750 BAKER ET AL Ann Thorac Surg STROKE AFTER CABG 2005;80:1746 50 insult in CABG surgery may be avoided is of high priority. We have identified the most consistent predictors of stroke to be older age, diabetes, and a history of stroke. These factors are not able to be manipulated. However, they provide insight into those patient groups at higher risk who may benefit from consideration of new and alternate surgical techniques, including off-pump and minimally invasive approaches and new instruments, including anastomotic and endoluminal devices, and monitoring such as the routine application of epiaortic scanning. This analysis was supported by the National Heart Foundation, Grant No. G 02A 0600. References 1. Bucerius J, Gummert JF, Borger M, et al. Stroke after cardiac surgery: a risk factor analysis of 16,184 consecutive adult patients. Ann Thorac Surg 2003;75:472 8. 2. Neurological outcomes and cardiopulmonary temperature: a clinical review. J Cardiovasc Surg 1995;10(suppl)475 80. 3. Salazar JD, Wityk RJ, Grega MA, et al. Stroke after cardiac surgery: short- and long-term outcomes. Ann Thorac Surg 2001; 72:1195 202. 4. Stamou SC, Hill PC, Dangas G, et al. Stroke after coronary artery bypass: incidence, predictors, and clinical outcome. Stroke 2001;32:1508 13. 5. van Wermeskerken GK, Lardenoye JH, Hill SE, et al. Intraoperative physiologic variables and outcome in cardiac surgery: Part II. Neurologic outcome. Ann Thorac Surg 2000;69: 1077 83. 6. McKhann GM, Grega MA, Borowicz LM Jr, et al. Encephalopathy and stroke after coronary artery bypass grafting. Arch Neurol 2002;59:1422 8. 7. Engelman RM, Pleet AB, Rousou JA, et al. Does cardiopulmonary bypass temperature correlate with postoperative central nervous system dysfunction? J Cardiovasc Surg 1995; 10(suppl)493 7. 8. Grigore AM, Mathew J, Grocott HP, et al. Anesthesiology 2001;95:1110 9. 9. Warm Heart Investigators. Randomised trial of normothermic versus hypothermic coronary bypass surgery. Lancet 1994;343:559 63. 10. Martin TD, Craver JM, Gott JP. Prospective, randomized trial of retrograde warm blood cardioplegia: myocardial benefit and neurological threat. Ann Thorac Surg 1994;57:298 304. 11. Mora CT, Henson MB, Weintraub WS, et al. The effect of temperature management during cardiopulmonary bypass on neuropsychologic outcomes in patients undergoing coronary revascularization. J Thorac Cardiovasc Surg 1996;112: 514 22. 12. Baker RA, Andrew MJ, Knight JL. Evaluation of neurologic assessment and outcomes in cardiac surgical patients. Semin Thorac Cardiovasc Surg 2001;13:149 57. 13. Antunes PE, Ferrao de Oliveira, J, Antunes MJ. Predictors of cerebrovascular events in patients subjected to isolated coronary surgery. The importance of aortic cross-clamping. Eur J Cardiothorac Surg 2003;23:328 33. 14. Borger MA, Ivanonv J, Weisel RD, Rao V, Penisoton CM. Stroke during coronary bypass surgery: principal role of cerebral macroemboli. Eur J Cardiothorac Surg 2001;19:627 32. 15. McKhann GM, Golsdborough MA, Borowicz LM, et al. Predictors of stroke risk in coronary artery bypass patients. Ann Thorac Surg 1997;63:516 21. 16. Naylor AR, Mehta Z, Rothwell PM, Bell PRF. Carotid artery disease and stroke during coronary artery bypass: a critical review of the literature. Eur J Vasc Endovasc Surg 2002;23: 283 94. 17. Andrew MJ, Baker RA, Kneebone AC, Knight JL. Neuropsychological dysfunction after minimally invasive direct coronary artery bypass grafting. Ann Thorac Surg 1998;66:1611 7. 18. Newland RF, Baker RA, Mazzone AL, Ottens J, Sanderson AJ. Removal of glucose from the cardiopulmonary bypass prime: a prospective clinical audit. J Extra Corpor Technol 2004;36:240 4. 19. Okano N, Owada R, Fujita N, Kadoi Y, Saito S, Goto F. Cerebral oxygenation is better during mild hypothermic than normothermic cardiopulmonary bypass. Can J Anaesth 2000;47:131 6. 20. Shaaban Ali M, Harmer M, Latto IP. Jugular bulb oximetry during cardiac surgery. Anaesthesia 2001;56:24 37. 21. Trehan N, Mishra M, Sharma OP, Mishra A, Kasliwal RR. Further reduction in stroke after off-pump coronary artery bypass grafting: a 10-year experience. Ann Thorac Surg 2001;72:1026 32. 22. Hirotani T, Kameda T, Kumamoto T, Shirota S, Yamano M. Stroke after coronary artery bypass grafting in patients with cerebrovascular disease. Ann Thorac Surg 2000;70:1571 6.